ML19289E437

Risk Assessment of Radioactive Waste Isolation in Deep Geologic Formations, from 781204-08 Meeting of Review Group for EPA

ML19289E437
Person / Time
Issue date:	12/08/1978
From:	NRC OFFICE OF STANDARDS DEVELOPMENT
To:
Shared Package
ML19289E435	List: ML19289E434 ML19289E437 ML19289E439
References
NUDOCS 7904180175
Download: ML19289E437 (55)

Text

.

9, a

O Report of the NRC Review Group for the EPA Risk Assessment of Radioactive Waste Isolation in Deep Geologic Formations - (Meeting of December 4 - December 8,1978) 7904I80115

(

Table of Contents Page 1.0 I n t ro d u c t i o n.................................................. 1 1.1 0 b j e cti v e s............................................... I 1.2 In fo rmatio n Ava i l abl e.................................... 2 1.3 Organi zation o f the Revi ew............................... 2 2.0 EPA Draft Standard and Its Use

~

o f t h e A DL Re p o rt...........................................

3 2.1 Th e S ub g ro u p ' s Fi n di n gs..................................

3 2.1.1 Development of a Standard Based on Ri s k Con s i de ra ti on s............................. 4 2.1. 2 Technical Basis for the Proposed EPA S t a n da rd........................................ 6 3.0 Review of Task D: Assessment o f Acci dental Pathways.......... 9 3.1 Th e S ub g ro up ' s Fi n di n gs.................................. 9 3.1.1 S e n s i ti v i ty....................................... 9 3.1.2 Un c e r t a i n ty...................................... 1 0 3.1. 3 Probabilis ti c Consi de rations..................... 11 3.1.3.1 S t re ams.................................. 1 1 3.1. 3.2 Canals..................................

11 3.1.3.3 Gas /0il Explo rati on Drilling............ 12 3.1. 3. 4 N u cl e a r Te s ts...........................

12 '

3.1.3.5 Volcanoes...............................

12 3.1. 4 Di s c re ti z a ti on................................... 12 3.1.5 Qual i ty o f Da ta Us e d............................. 13 3.1.5.1

........................................13 3.1. 5. 2

........................................13 3.1. 5. 3........................................

14 3.1.5.4 14 3.1. 6 Phys i cal Mo del i n g................................ 14 3.1.6.1 Vol ca ni c Even ts.........................

14 3.1. 6. 2 Faul tin g and Seismi ci ty.................

14

Table of Contents (Cont.)

P_ age 3.1. 6. 3 Failure Elements Involving Technology....

15 3.1.6.4 Water in Upper Aqui fe r................... 15 3.1.6.5 15 3.1.6.6 Screening................................

15 3.1. 7 Do c ume n t a ti o n..................................... 16 3.1. 8 Speci fi c Commen ts on S ub tas k D....................

16 4.0 Re vi ew o f Ta s k s C-1 an d C-2................................... 21 4.1 Th e S u b g ro u p ' s Fi n di n gs....................................... 21 Appendix 4-I Consequence Assessment Used by EPA................... 24 Appendix 4-II Governing Equation Used in ADL Report................

30 Appendix 4-III Speci fi c Commen ts on Tas ks C-1 and C-2.............

32 5.0 Re vi ew o f Ta s k s A an d B.......................................

35 5.1 Th e S ub g ro up 's Fi n di n gs..................................

35 5.1.1 S p e ci fi c Commen ts................................. 35 R e fe re n c e s......................................................... 4 0 Appendix...........................................................42 o Handout supplied to participants containing the main objectives, tentative agenda and suggested questions and areas to be included in the review.

o Follow-up letter to D. Egan, USEPA, from N. Trask, USGS.

it

List of Tables Page Table 3.1.1 Abbreviated Results of Computer Runs Made by EPA at Request of Review Group........ 10 Table I Parameter Values for Computer Runs Made by EPA at Request o f Review Group........ 26 Table II Computer Runs Made by EPA at Request o f Re vi ew Gro up.............................. 2 7 tii

Report of the NRC Review Group for the EPA Risk Assessment of Radioactive Waste Isolation in Deep Geologic Formations -

(Meeting of December 4 - December 8, 1978) 1.0 Introduction The Nuclear Regulatory Commission (NRC) through its regulatory and licensing actiona, has the responsibility to implement specific requirements of the U.S. Environmental Protection Agency (EPA) standard for high-level-waste isolation. There is, therefore, a need for early coordination between the two agencies.

Both EPA and NRC have funded risk assessment research programs in support of licensing criteria and standards development.

The NRC research work is being carried out at Sandia Laboratories, Albuquerque, New Mexico, and the EPA work was done by Arthur D. Little, Inc. (ADL), New Jersey. The NRC work covers a larger scope and time period than did the ADL work. There have been technical exchanges betweentg3 projects since inception of the ADL program., The EPA requested the Fuel Cycle Section of the Probabilistic Analysis Staff (PAS) to review the draft ADL report,and a brief informal review was done. However, it became apparent that an in-depth technical review was warranted since even partial attainment of the objectives listed below would be valuable to the NRC's mission.

Therefore, a Review Group comprised of NRC personnel and a selected group of consultants was established.

1.1 Objectives The objectives of the review were as follows:

(a) To identify significant weaknesses in the ADL work and recommend further work to correct any deficiencies. This would assist NRC in the direction and management of research projects in similar areas, in addition to assisting EPA.

(b) To contribute to the understanding of the capabilities and limitations of probabilistic risk assessment techniques with respect to waste isolation in deep geologic media.

(c) To examine the manner in which the ADL work was utilized to formulate the EPA draft standard and the degree to which the ADL results support that standard. One product of this examination could be insights into the use of probabi-listic risk assessment to support the formulation of standards for nuclear waste isolation.

. 1.2 Information Available The infomation avaigle to the Review Group consisted of the 4

ADL report, reviews by four consultants of the ADL report section entitled " Task D--Assgssment of Accidental Pathways,"

and other reviewers' comments collected previously by EPA and sent to ADL. In addition, the Review Group was briefed by Daniel J. Egan and from a session during which EPA and members of the ADL technical staff answered questions. The Review Group was supplied with an outline of the main objectives, the agenda, names of the participants and suggested questions be addressed in the review (see appendix).

1.3 Organization of the Review The Review Group met for one week starting on Monday, December 4,1978, in the Maryland National Bank Building, 7735 Old Georgetown Road, Bethesda, Maryland. The agenda followed closely the tentative agenda included in the appendix. _

The participants in the meeting may be divided into two categories:

those who were members of one of the subgroups that were formed to review specific portions of the ADL work and were actively involved in writing this report, and those who participated in the general meetings. All participants are listed in the appendix, and the members of the various subgroups are listed at the beginning of each chapter.

The Chairman of the Review Group was I. Craig Roberts, Assistant Director for Site and Health Standards, Office of Standards Development; the Technical Secretary was Michael C. Callingford, Section Head, Fuel Cycle Risk Application, Probabilistic Analysis Staff, Office of Nuclear Regulatory Research.

The organization cf this report in relation to the ADL report subtasks, the objectives of the subgroup responsible for reviewing specific portions of the ADL report, the chapter titles and the subgroup membership are shown prior to each of the next four chapters describing the results of our review.

. 2.0 EPA Draft Standard and Its Use of the ADL Report The subgroup charged with review of the EPA standard and its use of the ADL report received a briefing by Daniel J. Egan on the formulation of the EPA draft standard and the ADI report.

The objective of the subgroup was to examine the manner in which the ADL work was utilized to formulate the EPA draft stands-d and the degree to which the ADL results support the standard.

Subgroup Members:

G. E. Apostolakis (Chairperson), University of California, Los Angeles (UCLA)

J. E. Campbell, Sandia Laboratories L. A. Casey, NRC J. C. Helton, Arizona State University F. J. Pearson, U.S. Geological Survey (USGS)

R. B. Hofmann*

• Part-time member 2.1 The Subgrouo's Findings In establishing the draft environmental standard for waste disposal, it is our perception that the EPA proceeded through the following steps:

o Based both on the ADL draft report and EPA consequence analysis, complementary cumulative probability distributions were derived for the number of latent cancer fatalities in the first 10,000 years after closure of EPA's hypothetical repository.

The resultant risk, expressed as the expected number of latent cancer fatalities (about 150) was apparently considered acceptabl e.

o The contributions of the more important radionuclides

• to the risk were examined.

o A' set of representative radionuclides was selected by considering:

the radionuclides making the largest contributions to the

risk,

• Radionuclide importance can be a function of the hypothetical site and of the event chains considered.

. the mode of release of the more important radionuclides (i.e., whether through groundwater or by direct ejection to the atmosphere),

different chemical species in the radionuclide inventory.

A standard was then established for wgt wg congerep37 gresentative set of radionuclides (

C, Tc, I,

Np, Pu). This standard specifies permissible quantities and postulates probabilities for environmental release during a projected 10,000-year period after repository closure.

The remainder of this chapter is divided into two parts.

The first part consists of a general discussion and comments on the development and the particular form of a standard based on risk considerations. The second part contains discussions and specific comments on the manner in which the proposed standard is supported.

2.1.1 Development of a Standard Based on Risk Considerations The attempt to use risk analysis to establish environmental standards is commendable; this approach directs attention to the areas of the greatest significance. However, a standard based on risk analysis can take many different forms.

For example, it could be expressed in terms of expected life shortening for an individual of an assumed population. As a second example, the standard could specify permissible radionuclide concentrations in groundwater and/or surface water.

It is important to note that a standard could be supported by risk analysis but need not be explicitly probabilitic.

The expression of risk as the product of probability and consequences.

(P*C) in this standard has the following useful aspect :

it allows separation of the analysis of risk into different disciplines so that it can be approached from many perspectives.

This is important because of the type of review the standard will receive (e.g., public forum).

The establishment of a standard by an agency is the expression of a definite decision on its part.

Ideally, this decision should be the result of an analysis of the costs and benefits of a number of possible choices. The nuclear waste management issue is, however, enormously complex, and the decisionmaker must limit the analysis to only a few relevant factors. Therefore, it is not surprising that there is disagreement among experts as to the proper and workable way to set a standard that includes risk considerations.

We have the following concerns:

1.

Possible alternative forms of environmental standards should be systematically surveyed and presented, and the selected form should be justified.

During the many reviews the draft standard will receive, the numerical values in the standard will undoubtedly be questioned many times and the final values may be quite different from present values. Thus, at this preliminary stage, the fact that the standard is expressed in terms of probabilities and environmental discharge rates is more important than the numerical values.

2.

The form of the standard and the process used in its selection should be subjected to review independently of the numerical values used in the standard.

3.

In developing candidate forms of a standard and in deriving numerical values, alternative technologies and repository sites should be examined. Because of the great diversity of the geologic, geohydrologig and surficial characteristics of different sites, examination of only one site may not insure that the standard is sufficiently comprehensive.

~

4.

Should the standard be based on point values, (e.g., the expected risk), or should it be more detailed, such as'in the form of a probability vs. consequences curve?

5.

The concept of probability may not be understood in the same way by all persons concerned. Whether one is a Bayesian or a frequentist would make a difference. Thus, the question of how probabilities are defined must be addressed.

6.

How should the uncertainties be treated? In particular, if a P*C curve is to be established, how would the probabilites be determined? If a confidence level is specified, the question of interpretation of probabilities becomes even more crucial, since Bayesians and frequentists interpret confidence intervals di f ferently.

7.

Since the probabilities of rare events must usually be calculated, should the standard go as far as to specify the procedures by which those probabilities should be derived? If so, would the standard be realistic and useful?

~

8.

The accessible human environment needs to be clearly defined

• i.e., how does one define the boundaries of the biosphere?

. 2.1. 2 Technical Basis for the Proposed EPA Standard There are two parts to the EPA risk calculations:

1.

Calculation of the event-chain probabilities, and 2.

Calcul

• consequences for any chain of events.

We understand that the methods of calculating event-chain probabilities and the event probabilities themselves are those given in Section D of the ADL report.

Comments on these probabilities and computational methods appear in Chapter 3 of this report.

The consequence model was developed by EPA and is not described in the ADL report.

It includes a waste-leach submodel, a nuclide-transport (by groundwater) submodel, an environmental pathways and health effects submodel, and a means for combining probabilities and consequences.

EPA's model was not examined in detail by this subgroup, but the nuclide-transport submodel is known to contain a conceptual error in that it considers the upper-limit rate of nuclide transport to be set by the superficial (Darcy) velocity of groundwater flow, rather than by the interstitial velocity (see Chapter 4_of_this report). _It could_well contain other errors.

It is, therefore, necessary that the consequence model be reviewed for:

(1) conceptual validity and adherence to accepted physical laws; (2) correctness of the mathematical techniques used for its solution; and (3) freedom from computer programming errors.

For purposes of examining the model's technical adequacy, intermediate results of its calculations should be readily available. For example, Section C-2 of the ADL report presents two models for radio-nuclide transport by groundwater. Results of operating the EPA

- ~ ~ '~ ~ ~

radionuclide-transport submodel should be, but have not been, tested against results obtained by operating the ADL and other models on physically identical problems. Similarly, the results of other EPA submodels should be compared with those of other models that simulate the same phenomena.

The following types of data are required for operati,ng the EPA P*C model:

o Probabilities of events and event chains are taken directly from Section D of the ADL report. As comments elsewhere

. show, there is considerable doubt about the values, and their ranges, of the individual event probabilities, and there are errors in the method presented (and used by EPA) for combining these into event chain probabilities, o Values of the physical parameters required in the transport model include:

site geometry, hydraulic conductivity and gradient, porosity, and nuclide retardation factors, waste-leach rate, and fractions of repository affected by the several failure event chains.

These values are taken variously from Tasks B, C, and D of the ADL report; comments on the values appear in the discussion of those sections.

o The submodel by which nuclide concentrations were converted to health effects was not considered.

Comments on Deficiencies in Consequence Calculations The EPA model is unchecked and unverified, and, based on the ADL report results, it may be inadequate.

For example, the model in ADL Section C-2 shows the physical unreasonableness of ignoring nuclide-sclubility limits. The EPA trans~ port submodel is not sol ubility-limited.

Credibility of the ADL modcls or the EPA standard is not possible without a rigorous treatment of uncertainty. The absence of any uncertainty analyses and only a limited sensitivity analysis in the ADL work are major weaknesses. Examples of judgments that cannot be made because of the absence _of uncertainty analysis are:

EPA informally made the judgment that the P*C curves derived o

from its analysis indicated " low" health effects. The absence of an uncertainty analysis does not allow justification of the

~~

belief that the model is realistic. Therefore, actions or decisions based on the model results are questionable.

o The selection of representative or the "most hazardous" radionuclides to regulate is again unjustifiable because of the absence of uncertainty analysis. There is reason to believe that the uncertainty, if evaluated, would be large enough to substantiate other radionuclides being hazardous

' given other sequences of events or different quantities of source radionculides. The standard purportedly is to apply to all sequences relative to geologic disposal as well as alternative methods of disposal.

. 3.0 Review of Task D: Assessment of Accidental Pathways The subgroup which reviewed Task D of the ADL report was supplied with Task D, ADL Report Volume I, "Assessmegt AccMental Pathways," and letters from four consultants 8 The group was briefed ray Daniel J. Egan on the fomulation of the EPA draft standard The objective of the subgroup was to examine the evaluation of accidental events and their consequences affecting waste isolation over long time periods as well as specific problems associated with performing risk estimates and with the assessment of acceptability of risks. The subgroup was also to examine diRerences in analyses for different geologic media.

Subgroup Members:

R. L. Iman (Chairperson), Sandia Laboratories G. E. Apos tolakis, UCLA*

D. J. Fehringer, NRC*

R. B. Hofmann, NRC*

M. Johnson, Los Alamos Scientific Laboratory E. O' Donnell, NRC C. B. Oh, NRC*

W. R. Pearson, USGS*

D. S. Rubinstein, NRC*

N. Trask, USGS

• Part-time members 3.1 The Subgroup's Findings 3.1.1.

Sensitivi ty If standards to be set by EPA are to be related to a P*C curve in any sense, then some careful consideration must be given to the factors that detemine the placement of the high-low P*C curves in the P*C plane. It is obvious that assumed probabilities associated with initiating events have a great deal of influence

~

on the P*C values. However, the selection of specific values of input parameters should not be disregarded, as changes in these input values can easily be demonstrated to have an ot der-of-magnitude influence on the P*C curve.

As a' example, n

consider the following results obtained from the EPA computer code. The results that follow were all based on high-value (conservative) probabilities that were held constant from run to run.

There fore,

. changes in the output are due only to changes in the input parameters.

All retardation factors (RFs) either remained the same or were lowered by one or two orders of magnitude, as indicated in Table 3.1-1.

The output reflects the expected number of health effects, calculated as y P C.

$9 Table 3.1-1 Abbreviated Results of Comouter Runs Made by EPA at Request of Review Group SPC Velocity (ft/yr) 1 $I Description Horizontal Vertical RF RF*10-I RF*10-2 Base Case 60 100 71 164 4100 Base Case *10-2 0.6 1.0 40 These cases I

Base Case *10 600 1000 165 Variations in expected health effects shown in Table 3.1-1 can be seen to cover two orders of magnitude from 40 to 4100. This demonstrates the problems that can arise when an analysis is based on

" typical" values for input variables. A complete description of the cases calculated to generate the results of Table 3.1-1 is given in Appendix 4-I of Chapter 4.

The point is that these typical values have to be determined by " experts" in the field of hydrology, and there will be disagreement from one expert to the next. A thorough analysis should demonstrate this variability.

3.1.2 UnceFtiinty~

The conceot of uncertainty in the computed probabilities is handled

~

inadequately in the ADL report. The only indication of uncertainty given is an operational one; i.e., the report includes high and low values to probabilities of certain events. Guidelines for the inter-pretation of these values are conspicuously absent.

Inadequacy of the data should not preclude a more conventional confidence-limit approach any more than the novel high and low value limits provided here. Very wide confidence limits, if provided, are preferable to high and low value limits that are not defined. The difficulty in applying the methodology can be seen in the following two schemes given by ADL for computation:

. (a) Compute the high and low probabilities separately by adjusting parameters in an equation.

(b) Given the low probability, the high probability is taken to be an order of magnitude larger.

The arbitrariness of this approach is not always acknowledged.

3.1.3 Probabilistic Considerations These comments concern some of the probabilistic assumptions used in Section 5 (pp.63-205) of Task D of the ADL report.

~

~

~

~

~

The basic problem is the fol10 wing questionable assumption:

certain properties of interest are modeled as random variables that are distributed uniformly over the area of the contiguous United States.

The particular examples cited subsequently include the location of potential repository sites with respect to streams, canals, nuclear weapons tests, volcanoes, and exploratory drilling. These assumptions must be justified for this section of the report to be technically credible.

3.1.3.1 Streams (fill) '

The average distance between a repository and the nearest stream is taken as 20 km. The rationale for this value is given in Appendix D-XII of the ADL report.

Several problems in consistency occur there.

First, the value of 20 km is estimated using a Monte Carlo experiment on the western half of the United States. A dubious " check" on this experiment is based on considering the entire contiguous United States. Actual streams and lakes are replaced by 250 standard streams (240 km x 200 m x 3 m) placed in parallel. More justification for these assumptions must be provided.

Furthermore, what are the implications of the model in terms of the high-value results?

Second, the role of climatic change, geomorphologic processes, and their possible variation with time over the 10,000- or 1,000,000-year time period was not considered.

Reference should be mad to work by Leopold, Schumm, Hack, Maddock and others with the USGS.g-Il 3.1.3.2 Canals (pp. 142-143)

The unrealistic assumption is that 2000 km of canals can be replaced by 40 canals (each 50 km long) distributed uniformly over the contiguous United States. The final ADL report should address the impacts of these assumptions on the high value computations.

. 3.1.3.3 Gas and Oil Exoloration Drilling (pp. 143-146)

The distribution of exploratory drilling holes is, of course, not uniformly distributed over the entire contiguous United States.

This exploratory drilling is selectively performed in only certain regions. Again, the high value seems questionable on these grounds.

3.1.3.4 Nuclear Tests (p.141)

The idea that nuclear test sites (after a 100 year moratorium) would be located randomly throughout the United States is incredible.

3.1.3.5 Volcanoes (p. 118)

Volcanoes are not located uniformly throughout the United States, and volcanic activity rates can change. Some form of regionalization should be used. The region should be defined by areas of comparable volcanic activity, surface-water distribution, climate patterns, and mineral resource potential (see appendix,

! tam 2).

3.1. 4 Dis cretization Section D-4.1.3.

The entire section should be rewritten to present the methodology in a clear way, especially when concepts (e.g., the treatment c' conditional events and the concept of consequential events) that are not standard are used.

The discretization of the time span (0,T) into 10 intervals should be fully explained and justified.

In the present formulation, the accuracy of the results depends on the numerical value of T.

While the model may well be conservative, the degree of conservatism would be different if T were 10,000 or 1,000,000 years,and a quantitative measure of this conservatism should be derived. The numerical example of Section D-4.2.3 used T = 20 years, which is very small considering the mean times to occurrence of the events A and C (1,000 and 10,000 years, respectively). This example should also be done using the exact expression for the probability that A, B, and C will occur (without discretization). Then the exact results should be compared with the results from discretization for a range of values for T.

The example, then, would be useful in demonstrating the accuracy and validity of the discretization technique.

. The equation at the bottom of page 29 needs justification.

It is not evident how the probabilities PI7 7 3 3, etc., would be determined when dependencies exist. This equattoR does not handle conditional probabilities in the standard way.

The two equations at the bottom of page 30 are incorrect. Let us take the following numerical values:

n = 2, N = 2, P ) = PI2 = 0.90 I

Then P(II, I2, 2) = P ) PI2 (22 - 1) = 0.92 1

x 3 = 2.43, a probability greater than unity.

3.1.5 Quality of Data Used The analysis uses the best data available in most places. Much of it has large uncertainties that should be acknowledged.

In many cases, bounding values are sufficient and should be so identified.

3.1.5.1 The data used for the probabilities associated with volcanic activity (D-5.4.3) date from 1960 and are outmoded. Up-to-date data are available through the geothermal program of the USGS (USGS Circular 726, for example). Nevertheless, little change in the volcanic probabilities would result.

3.1.5.2 The approach to estimating fault probability (D-5.4.4) apparently is to survey measured values of fault density over the country and take the average; if so, this should be explicitly stated (p. 134). The quality of the data used for this calculation cannot be assessed because individual sources are not given. These sources of data should be provided. The relevance of fault data from stable areas such as western Massachusetts is questionable since most of the faulting ceased at least 100 million years ago.

Data on faulting rates for the last 10 to 15 million years would he more relevant and would provide a meaningful prediction for the next 10,000 years. Preliminary data of the so used in deriving faulting rates are available.g that could be Different fault data and a different fault model will have a significant effect on the outcome, especially for early time periods.

. 3.1.5.3 The content and tone of the section on shaft and borehole sealing (p.148) are at odds with the widespread co technology for plugging does not nowgxist.gensus that a demons Experience with other seals in geotechnical projects indicates that estimating the perfomance of engineered works over periods such as a million years is highly uncertain.

Failure of tha seal itself must be given a nonzero probability over the entire 10,000-year period of analysis at least until a demonstrated technology exists.

The consequences of shaft seal failure will depend on the vertica]3 groundwater flow and can be minimized by suitable site selection.

3.1.5.4 Data in Section D-5.5 (long-tem degradation) is taken in many instances from outdated elementary textbooks and articles. Signi ficant changes in the data are not anticgated,but updated references are essential.

For salt-dome growth, see Bulletin of the Geological Society of America,1972.

3.1. 6 Physical Modeling 3.1.6.1 Volcanic Events - D-5.4.3 The model would he improved by eliminating from consideration the active plate margin of the continent. This reduces both the number of vents and the area considered. New compilations of vents will still yield a higher number of vents (>10) for the remaining United States, although the effect of this new data on total risk will probably be negligible.

3.1.6.2 Faulting and Seismicity - D-5.4.4 As noted above, the only valid way to work this problem is by actual analysis of specific sites to evaluate the rate of fault development.

There are many difficulties with dating faults and fault movement, but progress is being made. A generic estimate that is somehow representative of the United States as a whole is virtually meaningless. Recurrence rates for relatively inactive faults are slowly being compiled.

One fault in New Mexico was estimated to have a recurrence rate of approximately 100,000 years.15 However, it is not clear what the rate of migration of tectonic domains with associated faulting rates may be. The assumption of rate consistency over 100,000 years is unsupported. Careful site selection should yield a relatively low faulting rate, but the uncertainty will be great.

. 3.1.6.3 Failure Elements Involving Technolooy Analyses of the effects of heat are largely in the appendices and have not been thoroughly reviewed. The corrosion rates in Section D-IV are inapplicable because they refer to sodium chloride solution and seawater--not the strong brines or bitterns encountered in bedded salt.

The discussion of brine migration is over simplified. After the repository is sealed, fluids in the vicinity of the canisters will be brines, not steam, with a potential to hydrofracture or to cause loss of strength. These deficiencies are moot because the NRC will formulate rules to insure that the probabilities from such failures are acceptably low. Examples are requiring the shaft to be located away from the major heat sources and limiting the temperature of the waste.

One self-induced failure that may be difficult to model, and hence to provide for in rulemaking, is the far-field deformation of the upper barrier.

For spent fuel, the maximum far-field stresses will occur 1,000 years after closure, not 100 as implied in Section D-5.4.6.4 and D-5.4.6.5.

Feedback effects between cracking and solutioning (runaway solution) should also be adh.d under this heading.

Failure probabilities could be highee than showr.

3.1. 6. 4 Water in UpoerJquifer (pp. 155-156)

Estimates of this probability that are not equal to unity are unrealistic, as noted by Mr. Egan. Assigning the value of unity will have only a slight effect on the risk.

3.1.6.5 The fraction of the repository aff tcted, reprennted by the factor f, has been given valces that appear o be low. Under some circumstances, large volumes of salt can dissol4e in breccia pipes; thus, runaway dissolution need careful Waluation before the low f-values according to the extt:nt of salt disolution_can be use ful.

3.1.6.6 Screening Screening of event sequences or other approaches that could allow a risk contribution basis for selection of those scenarios treated

. in the report is not evident. Therefore, the basis for the choice of scenarios and event chains that were treated is not quantitative.

An example of this lack of completeness is illustrated in the following scenario not included in the ADL report. One possible scenario for human intrusion would be the development of a pumping ground-water well in an aquifer that is in hydraulic connection with leaky hydrostratigraphic units that surround the repository. This scenario would be characteristic for long-term assessments where governmental supervision is not assumed. Moreover, the point of human contact could be all nearby aquifers, and not just the overlying "uoper aqui fer(s)" and surface discharge points.

3.1.7 Documentation In order to assess how meaningful or useful the probabilities and failure rates developed in Task D are, complete documentation should be provided.

For example, where did the data come from, how were they derived, and how reliable were they?

The EPA computer code (Section D-4.2) needs to be documented further.

It should either be included as an appendix to the ADL report or published separately. Section D-4.2 should address the question of uncertainty. Was an uncertainty analysis performed?

~

If so, details and results of the analysis should be provided.

3,1. 8 Specific Comments on Subtask 0 1.

Page 2, item 2 - Needs to be reworoed.

2.

Page 3, fifth line from bottem - The phrase "hich and low" needs to be defined. quantitatively if possible.

3.

Page 5 - The figure on this page is not meaningful without uncertainty bounds.

4.

Page 6, first item, last three lines - The last word, " eventually,"

seems to be unnecessary. Please explain its use more fully.

5.

Page 8, top of page - This statement is in error. Explain how it was derived.

6.

Page 18, first two sentences - These sentences are too inclusive and need to be rewo_rded.

7.

Page 23, first item, first sentence - The word " calculating" is unnecessary; change the word "possible" to " considered."

. 6 8.

Page 25, line 11 - Is the 10 -year time span used consistently.

throughout the report?

9.

Page 24, fourth line from the bottom - The word " changes" should be

" chances."

10.

Page 26, figure D This figure needs a better explanation.

11. Page 28, line 11 - Failure rates should determine the number of intervals.

12. Page 30 - Reference is made to Equation (1),which cannot be found anywhere.

13. Page 31 - Elaborate on item 3.

Does it imply that a cor.non-mode failure analysis was not done?

14.

Page 31, item 4 - The expression is explained incorrectly.

It is not a conditional probability.

15.

Page 32, first paragraph - This explanation is wrong.

16. Page 36, first new paragraph - Please define " time phase" and how it is used with " time interval."

17.

Page 37, second paragraph, line 5 - This probability should not be zero.

5

18. Page 37, fifth line from bottom - Compare answers if 10 years is used instead of 20.

19. Page 38, line 4 - What " data values"?

20. Page 39, first new paragraph - P should be the probability of HA "A" not healing.

21. Page 40, item 2 - What is meant by " correlate"?

22. Page 46. Table D The footnote is incorrect.

The charge to the reactor should be 50,000 MTHM.

23. Page 50, line 50 - No figure number given.

24.

Page 71, line 13 - No figure number given.

25. Page 104, figure 0 This figure is not understandable. What are the data that led to this figure?

26. Page 105, equations 21 and 22 - The probability goes to infinity as the meteorite diameter goes to zero.

Please explain.

. 27. Page 112, line 12 - Fill in missing equations.

28. Page 115, second paragraph, lines 5 and 12 - The references are out of date and inappropriate.

29 Page 118, second paragraph - This paragraph is wrong. Even if the paragraph were right, the underlying doctrine is debatable.

30.

Page 126 - The word " flaw" should be replaced with the accepted geologic term.

31.

Page 129, Figure D What are the units of length? " Repository" is misspelled in the caption.

32.

Page 130 - In the eastern United States there may not be a relationship between seismicity and faulting.

33. Page 134, last line - This line is incorrect. A pre-Tertiary event would be at least 60 million years old. An early Tertiary event could be 20 million years younger.

34. Page 138, first new paragraph - This discussion is a contradiction.

Low-grade salt is not a suitable site for a repository due to the presence of water-bearing clay.

35. Page 140, Section D-5.4.5.3 - This section is not convincing.

What does dam sabotage have to do with nuclear repository sabotage?

Page 142, Equation 15 - This equation is dimensionally incorrect.

36.

37. Page 160, line 17 - Contrary to the statement, solution features in Florida, Ohio, Kentucky, Indiana, Jamaica, and Puerto Rico often have structural control.

38. Page 160, line 19 - Reference 67 is incomplete.

39. Page 165, line 23 - Reference 1 is not appropriate. The original source material should be cited.

40. Page 168, figure D-30a - What is the original source of this diagram?

41. Page 170, figure D The drawing should be explained by a complete caption. A scale should be provided.

42. Page 171, figure D What was the source of this figure? A complete caption and explanation would make it more meaningful.

. 43. Page 175, line 2 - The word " discussed" is used loosely here. The material (p.163) was only briefly described.

44. Page 175, line 11 - Geologists are predicting another glaciation within 10,000 years, but not with great certainty.

45. Page 176, Section D-5.5.3.2, first paragraph, line 6 - Check the validity of the number 1.5x10 6

46. Page 177, figure D The figure should be captioned and described.

47. Page 179, line 19 - The reference is inappropriate.

Page 179, line 21 - The reference by Ewing and Ewing is about Q-48.

years old. A more recent reference should be used. See Gera.

49.

Page 197 - This table is not legible.

50. Page 199, line 2 - Reference is missing.

51.

Page 253 - The word "fomat See Code of Stratigraphic Nomenclature.{gn" is misused.

0

52. Page 253, Section D In general. Section D-7 was poorly done. The extrapolations are not fully developed.

53. Page 253, sixth line from bottom - Reference is missing.

54. Page 255, line 5 - The phrase " highly metamorphic rocks" should

_ read " highly metamorphosed rocks."

55. Page 255, line 17 - Granite is not an igneous mineral.

56. Page 258, figure D This figure is intended to show granitic rocks, however, volcanics and granites are lumped together. Many surficial granites are omitted such as the St. Francois Mountains, the Arbuckles, the LLANO uplift, and the Wichita Mountains.

57. Page 259, line 12 - There has been petroleg production from fractures in igneous rocks. See levorsen 58.

Section 8 - General Comment: Breakdown of subsections like 8.i.j.k.l.m.n gets out of hand and defeats the purpose.

59.

Page 309 - What is meant by "significant event"? Please define this.

60.

Page 313 - Comment:

It is too much to expect fault-tree analysis to guarantee to identify all significant events. The analyst must adapt the methodology to his application. The methodology itself cannot be expected to provide identification of the pertinent events.

61.

Page 316, last sentence before 8.1.1.3.2 - Identification does not necessarily imply better precision in the risk estimates.

62. Page 319 - The statement " experts predict only slightly better than non-experts" is meaningless.

63. Page 321 - How was the table produced? Since "probabilistic" forecasting has such a low rating, why was this study done "probabilistically"?

64. Page 332, first paragraph - How was it concluded that solar energy was a low-risk sogce? Other experts disagree; for example, recent articles in the British periodical New Scientist (1978) indicate otherwise.

65. Page 333 - This discussion should be expanded to clarify its relation to risk assessment.

66. Page 351, Section D-8.2.3 - The review of the work on risk acceptage is not complete; for example, the work by Okrent and Whipple should be included.

67.

Page 352, second paragraph

.In." definitive work " delete the word " definitive."

. 4.0 Review of Tasks C-1 and C-2 The subgroup that reviewed Task C of the ADL report was supplied with the Task C-1 Report, " Assessment of Geologic Site Selection Fcctors,"

and the Task C-2 Report, " Analysis of Migration Potential." It was also briefed by Daniel J. Egan on the formulation of the EPA draft standard.

The objective of the subgroup was the assessment of geologic and hydrologic site-selection factors, including associated model development and analysis of migration potential.

Subgroup Members:

J. J. Curry (Chairperson), NRC G. D. DeBuchannane, USGS F. A. Donath, University of Illinois C. D. Hollister, Woods Hole Oceanographic Institute

• M. E. Krug, NRC R. B. Lantz (Informational Consultant), Intera Environmental Consultants, Inc.

T. J. Nicholson, NRC F. J. Pearson, USGS*

J. E. Campbell, Sandia Laboratories S. F. Schreurs, NRC

• Part-time member 4.1 The Subgroup's Findings When the subgroup began its review of the ADL work, the extent to which the geotransport modeling presented in the ADL report was used by EPA in its consequence analysis was not clear.

Furthermore, different models we.e used in the ADL report for various purposes. Some investigation led to an understanding that the model used by EPA to calculate consequences for its P*C analyses was a simplified version of the one-dimensional steady-state ADL model for radionuclide transport in porous media. As presented in Appendix 4-I to this chapter, the governing equation for the EPA model does not include decay chains, dispersion, dilution, temperature,and solubility considerations.

The governing equation of the ADL model, used in Appendix D-XIII of the ADL report is presented for comparison in Appendix 4-II of this chapter.

Major conclusions and recommendations of the subgroup's reviaw of Tasks C-1 and C-2 follow.

Detailed comments on these tasks are included in Appendix 4-III of this. chapter.

. In the ADL analysis, the source term for radionuclide transport was limited by the leachability of the waste form. In response to questioning by the subgroup, it was stated that there was no overriding technical reason for choosing this limitation on the radionuclide source term. An alternative approach is to also limit the source term according to the solubility of the radionuclides in the transporting fl ui d.

The inadequacy of assuming only a leachability source tem limitation is apparent when one considers that such a basis for analysis could result in a prediction of higher radionuclide releases for some radionuclides (e.g., U, Pu, Am, Np) than would result from a solubility source-term limitation. A leachability type of limitation would perhaps be acceptable for a " conservative" (i.e.,

higher than actual health effects) analysis of a spect fic site,as might be used for a particular licensing application.

However, the subgroup questions this approach as a technical basis for setting standards for high-level-waste repositories because it does not necessarily reflect a risk that would be as low as might actually exist under realistic conditions.

Furthermore, it should be noted that the ADL report typically utilizes high values for the probability of initiating events and " conservative" values for the parameters that determine the consequences of those events. This is likely to result in estimates of releases that are higher than would actually occur.

It should be emphasized again that the use of such high estimates of releases in standard setting could restit in "nonconservative" standards because of the unrealistic assessment of risk under actual conditions.

As a rt ference document for use as the basis for developing radiation standards relating to high-level-waste management, Tasks C-1 and C-2 must withstand the scrutiny of technical review.

In its present draft forn, these sections lack credibility.

For example, the discussf or of distribution coefficients does not reference work performed in the last five years.

The discu:sfon of plate tectonics is elementary and not comprehensive.

Further, the use of references to personal corrmunications with persons other than widely recognized experts does not add to the stature of the report.

It is recommended that all subjects, if judged appropriate for inicusion in the report, be accurately and thoroughly addressed.

The subgroup concluded that the sensitivity analysis provided in Tasks C-1 and C-2 is neither adequate nor supportive of the EPA s tandard.

This conclusion is based on the following:

1.

The approach used for the sensitivity analysis was not presented.

It appears that the sensitivity analysis is only a limited parametric study.

. 2.

Two migration models, one analytical a..d one numerical, were used to investigate different parameters. The rationale for the use of the two models was not presented.

3.

The isotopes icentified as most important in terms of health effects from the sensitivity analysis do not include the mog important isotope identified by the epa or other investigators (99 c).

T 4.

The parametric analysis in the report is limited to only the migration n.adeling. Specifically, there is no attempt to examine the sensitivity of risk to selected scenarios or to event probabilities.

It is therefore recommended that a comprehensive sensitivity analysis be perfoni;ed and thoroughly documented. _

. Appendix 4-I Consequence Assessment Used by EPA This appendix describes the results of modeling runs made by EPA. at the request of the review subgroup during the review session. The objective of those runs was to perform a simple parametric analysis on the EPA model.

The risk assessment used by EPA in its in-house analysis (based on our understanding without having written documentation) can be outlined as follows:

1.

It utilizes the probability of initiating events from Task D of the ADL report.

2.

It utilizes the fraction of the repository affected by various initiating events developed from Task D of the ADL report.

3.

It utilizes parameter values (permeability, gradient, porosity, etc.) for pathways developed in Task C-1 and for the specific " worst probable" case from Task C-2 of the ADL report.

4.

It uses a simplified geosphere pathway model developed in house.

The critical parameters are leach rate (LR), initial quantity of the radionuclide of interest (Q ), fraction of the repository affected (f ),

g 9

time of interest (t),and the total delay time (t ) which is subdivided into:

a A.

lime of occurrence of the event, t e

B.

Travel time through the repository and communication area, which depends on:

(1) travel path distance through the repository and communication area, X(r-aq)

(2) interstitial velocity, V (r-aq) w (3) retardation factor, R (r-aq) d C.

Travel time through the aquifer to the surface water or

" accessible environment",which is similarly dependent on:

(1 ) travel path distance through the aquifer to the

" accessible environment," X (aq-sw)

(2) interstitial velocity, V (aq-sw) w

. (3) retardation factor, R (aq-sw) d The resulting governing equation is e (LR) (t-t )

-At Q (Ci/yr) = (LR) (Q ) (f ) e d

g g

whe re R (aq-sw)

R (r-aq), X(aq-sw)

X ( r-aq )

d d

td=te+

yw(r-aq)

VW(aq-sw)

The EPA in-house model neglected dispersion and daughter product formation (e.g., 226 a, which was identified as a potentially important health-effect R

nuclide in the ADL report).

In EPA's consequence assessment the parameters listed in Scenarios 1 and 2 of Task C-2 (worst probable case) were used.

For the two event pathways contributing the major portion of the health consequence (vertical pathway to surface water over the repository and horizontal pathways to a one-mile biosphere discharge point, as shown in Figure 1), the above referenced parameters were used.

Table 1 summarizes values for the parameters that the review group specified be used for comparison runs made during the review. These runs were made by EPA at the request of the review group and are listed together with the results in Table 2.

Although high and low values were included for the probabilities associated with initiating events, little sensitivity analysis was performed over a large range in parametric values in the actual risk assessment given the initiating event.

For example, there would still be a sizable uncertainty in permeability and porosity created by an initiating event such as a fault or breccia pipe. The review group requested that a very limited parametric analysis be performed, and EPA was able to provide results from this limited analysis. The following section summarizes the results and preliminary coments.

Figure 1 Simplified Geometry and Definitions

,,,,,,,,,, Ground Su r face V,R n

y d

One-mile horizontal path

$1500 ft V,R d

V V,Rd"I y

R Reposi to ry

TABLE l_

Parameter Values for Computer Runs Made by EPA at Request of Review Group Interstitial Permeability Velocity Pathway (cm/sec)

Po rosi ty Gra dient Reta rda tion (ft/yr)

-4 Horizontal 10 0.15 0.1 i:o rmal *

~60 Vertical 10-4 0.01 0.01

• 100 b

Salt and Interbeds None Aquifer to Surface No rmal*

• Refers to the retardation factors in Task C-1 of the ADL report.

TABLE 2 COMPUTER RUNS MADE BY EPA AT REQUEST OF REVIEW GROUP Ground wa ter llori zontal Vertical Expected Case No.

Des cription Velocity (ft/yr)

Velocity (ft/yr)

Retardation factor Health Effects 1

Base 60 100 "No rmal "

154 (See Table C-3 of ADL Report) 2 Base Modi fled R 60 100

" Normal" except for*

71 d

Tc = 10 C=1 g

3 Low Veloci ty.

0.6 1

Case 2 40 Modi fled R I

d 4

liigh Velocity 600 1000 Case 2 165 Modi fled R d S

Base, Modi fied, R d 60 100 (reduced by 10)

Case 2 times 1/10 164 Base, Modi fled, R 60 100 Case 2 times 1/100 4126 6

(reduced by 100) d

• In Table C-3 of ADL Task C-1, Tc = 1, C = 10

It should be noted that the nuclides causing the health effect can change substantially for the various cases. The top three nuclides for each case are summarized below. The numbers in parentheses represent health effects.

Total Case Nuclide Contribution Health Effects 1

99Tc (84), I4C (49), 237Np (17) 154 2

14C (50), 237Np (17), 241 m (3) 71 A

3 I4C (36), 241Am (3), 1291 (1) 40 90 r (76),14C(50),237Np(21) 165 4

S 5

90Sr(76),14C (50), 237Np (21) 164 6

239Pu (2520), 90Sr (1360), 241 m (73) 4126 A

The " normal" values for the retardation factor, R, included Tc with an R 4

d of unity. There is evidence

• that for the oxidation state in most ground-water, Tc will have a valence of +4 and as a cation will be considerably retarded. Apparently what has been assumed in the EPA analyses is that Tc will exist in the +7 state as an anion. Thus, little retention would occur.

Retardation factors for other nuclides can be substantially affected by salinity, pH, or other factors.

For example, the Kg for Cs can be reduced by a factor of 100 by increasing the salinity of the solution to sodium chloride saturation.**

The parametric analysis requested covers a range of 100 in retardation factors and specifically assigns Tc a high retardation factor.

The results of this limited parametric analysis show significant, as well as insignificant, changes in the expected value of health effects over 10,000 The effect of a change in velocity by a factor of 1,000 shows a change years.

in the expected consequence by a factor of 4.

Further, at the high-velocity value of groundwater (Case 4), 90Sr becomes the most significant nuclide.

If the " normal" retardation factors are high by a factor of 100, Pu becomes the

• Chemistry of Technetium and Neptunium in Contact with Underweathered Igneous Rocks, E.A. Bondietti and C.W. Francis (Environmental Sciences Division, Oak Ridge National Laboratory) (Prbgram Abstracts, Annual Meeting Materials Research Society, Boston, Massachusetts, November 1978).

• • USGS Water Resource Bulletin ll40-A.

. most important nuclide with Sr also important. Thus, for what could be termed a variation due to data uncertainty and site variability, the expected consequence could apparently vary by a factor of 100. The above comment applies only to an effect of data uncertainty and site variability.

It does not include the uncertainty in determining the probability of initiating events.

.. g

. t Appen' dix 4-II Governing Equation Used in ADL Report 7

(Example, Appendix D-XIII of the ADL report)

The flux K (Ci/sec) of material crossing the plane x is given by K(x,t)=nA(UG-h) e x + (Ut/a) 4((Dut )/a]'A

, [x - (Ut/a)]2 At }

3 4Dt/a a = retardation factor

• A = the area of flow 2

D = diffusion coefficient (cm /sec) exp = exponential function 3

G = the concentration (Ci/cm ) at any point in space, x, at any time, t, after release for an instantaneous l-curie release at time t=0 n = the effective porosity of the medium e

t = time U = groundwater velocity (cm/sec)*

x = path length (cm)

A = decay constant (sec-I)=

hal f-l fe There is inconsistency in the use of symbols throughout the ADL

~

report, e.g., "a" and "Rd," "n " and "Ne."

e

)

. The ADL report gives an example of a consequence calculation in Appendix D-XIII of Task D.

The example is for 129I. The variables used in the ADL report, for comparison with those used by EPA (see Appendix 4-I),were:

a = retardation factor

• 2 A = area of flow for horizontal transport (cm )

2 D = diffusivity (cm /sec) f = fraction of the repository involved with the flow K = permeability (cm/sec)

N = effective porosity e

V = groundwater velocity (cm/sec)**

x = path length for vertical transport (cm) r = tortuosity factor for fractured salt

• There is inconsistency in the use of "a" and "R " througNut the ADL report.

d

• • There was significant confusion throughout the report regarding the use of Darcy and interstitial velocities which was carried through the EPA analysis.

e

. __ Appendix 4-III. Specific Comments on Tasks C-1 and C-2 C-1 1.

Page 4 - Change phrase " levels of higher pressure to lower pressure" to " levels of higher fluid potential to lower fluid po ten tial. "

2.

Page 5 Table C-1 a.

Porosity should be " effective porosity".

b.

Low end range of porosity values seems high.

c.

Range of permeabilities is questionable, d.

Footnote noting existence of " caprock" above the dome salt should give either characteristic parameters or definition of " caprock".

3.

Page 6 -- a. Definition of "V" should be " Darcy velocity".

b.

Explain why turbulent flow "should not" have an effect on analysis of " suitable repository" (third paragraph).

4.

Page 7 - We disagree with statement regarding permeability that the actual value can be accurately measured only in closely controlled laboratory tests.

5.

Page 9 - a.__The quantity c should be defined as " effective porosity".

b.

Change definition of " sorption" to " delay" rather than " remove" nuclides.

6.

Pages 12, 14, 15 a.

Table C-3, Fig. C-1 and Table C-4--Kd values appear to be too high.

b.

Ref. 7 is limited source for data.

7.

Page 17 - Favorable waste-rock interactions were not noted, i.e.,

in terms of inhibiting radionuclide movement.

8.

Page 18 - a. In a description of waste-rock interactions in the last paragraph, vapor formation or hydrofracturing was not noted.

b.

Assumption that recrystallized minerals are stable under new P-T conditions is questioned, i.e.,

eutectic changes could result in lower melting points.

9.

Page 19 - Reference to Table 1 for thermal conducting values should be changed to Table C-1.

- 10. Page 21 - No mention of " mechanical properties" of rock made in

" Site Selection Factors".

11. Page 24 - At what P is second sentence true?

12. Page 61 - The age of water is not necessarily imoortant, rather its time of discharge.

C-2 1.

Page 2 - Define " upstream" and " downstream" in more conventional and appropriate terms.

2.

Page 3 - State that arrows in Figure C-8 do represent vectors and are for. illustration purposes only.

3.

Page 7 "U" is Darcy velocity, not interstitial velocity.

4.

Page 8 a.

We don't necessarily agree with broad statement that 25-1000C temperature increase will increase leachability by a factor of 10-100; what waste form, media?

b.

State assumptions on the time-dependent behavior of waste form.

5.

Page 9 - Change "is approached at longer times; however, much of the data..." to "is approached at longer times.

Much o f the data...."

6.

Page10 - Third paragraph is correct and should be expanded in terms of leach rate and solubility. We note that there is little good leach data available.

7.

Pagell - Remove dotted line from graph, and on ? age 10, fifth line, change ' extrapolation' to ' overlap:

8. Page 12, C-4.3, second paragraph - Note that although there is still a.hiah uncertainty factor for site-specific modeling, it is much less than that for generic modeling. However analyses have not yet been completed that show the extent of the differences in site-specific and generic modeling uncertainties.

9. Page 15. second paragraph

" Sorption" is too inclusive a term.

10. Page 16 - Expand discussion on numerical dispersion to include limitations on grid-block size and time step.

11. Page 17 - Qualify statement regarding " continuum" approach.

12. Page 21 - Change "nuclide stability" to "nuclide solubility."

13. Page 22. Table C-7

- In place of

• dispersion;* *dispersivity"would be a more acceptable term.

14. Page 24 - Clarify that T applies to peak arrival time.

15. Page 25 - Change footnote on bottom of page.

16. Page 50 - Change term " communication area" to " communication pathway."

17. Page 52 - We question need of Section C-4.5.

18. We suggest that EPA report EPA /520/4-78-004, " State of Geological Knowledge Regarding Potential Transport of High-Level Radioactive Waste from Deep Continental Repositories",

be used as a reference for model development.

. 5.0 Review of Tasks A and B The subgroup that reviewed Tasks A and B of the ADL report was supplied with the Task A Report, " Source Characterization / Definition", Appendix A, " Pertinent Nuclear Technology and Sources of Radioactivity", and the Task B Report, " Effectiveness of Engineering Controls" It was also briefed by D.J. Egan on the formulation of the EPA draft standard.

The objective of the subgroup was to review the state of the art for applying engineering controls to waste isolation. Assessment or measure of the overall effectiveness of engineering control and the rate at which radioactive material becomes available for transport out of the repository should take into account source-term characterization, repository design, and waste form performance.

Subgroup Members:

L. A. Casey (Chairperson), NRC M.E. Krug E. O'Donnell, NRC W.R. Pearson 5.1 The Subgroup's Findings The information presented in Tasks A and 3 is minimally related to the remainder of the report. This extraneous information contributes to confusion about identification of what data were used for the analyses done. The group suggests the next draft make Tasks A and B appendices to 1 different basic report. The basic report should be written such that the appropriate analysis sections of Tasks C and D are clearly related to the study objectives. Not only should the report sections be written to show their relationship to the study objective but also the rationale for selecting data and a summary of the input or source data used in the analyses should be stated explicitly.

5.1.1 Specific Comments 1.

Table A-II.1, p. A-II-5.

This taole should be revised to reflect the important nuclides identified in the ORIGEN runs.

2.

Section C-4.4.

The analysis for migration of radionuclides in the geosphere does not include important nuclides identified in the ORIGEN calculations which may contribute significantly to the accident dose (e.g.,99 c ).

Thus, Table C-9 appears to be T

incomplete (on comparison with Appendix A-IV).

(See Chapter 4 of this report.)

3.

Page 1, A-1.1, B.l.1, Paragraph 2, third sentence.

It may be that EPA will consider fuel cladding and associated hardware in the high-level-waste (HLW) category.

For the purpose of licensing by NRC under the Energy Reorganization Act of 1974, these materials are not HLW.

4.

Page 4, B-1.1, Paragraph 3, second and third senterces.

These sentences must be revised because ERDA is now DOE and the generic EIS has still not been issued.

5.

Page 5. B-1.2, Paragraph 2, third sentence.

Replace " test the sensitivity of" to " evaluate."

6 Page 8, B-2.2.1, thi rd paragraph.

Tnis paragraph should explain more fully what options are available and advantages and disadvantaces of each, if a preferred technology is suggested. Explain why they are preferred.

7.

Page 11, B-2.3, third paragraph.

Has it been decided not to consider reference cases based on foreign technology, or is it only for the purposes of this report (short time limit)?

8.

Page 12, B-2.3, fourth paragraph, last sentence.

There are three separate thoughts in this sentence, concerning spent-fuel leachability, canister corrosion, and radioactive-material transport.

It should be rewritten to clarify intent.

9.

Table B-1, page 13, topic Iodine-129.

In several places in Task B reference is made to storing mercuric iodate in steel containers. Unless an inert (plastic) liner is used, it will not stay there very long.

- 3 7-

10. Page 14, B-2.4, fi fth paragraph.

Use of the Y/0WI/TM-16/(22 volumes), Technical Support for GEIS:

Radioactive Waste in Geologic Formations, could supply valuable information in this area.

11. Page 15, B-2.4, seventh paragraph.

In the fifth sentence, concerning elevated temperatures, suggest that media conductivity (among other things) will affect peak temperature.

12. Page 16, B-2.4, second paragraph, sixth sentence.

Why is the planar heat density limited to 126 kW/ acre? OWI design suggested 150 kW/ acre as being sufficiently low for salt.

13.

Is there a reason that no " references" are used until Section B-3.27 It is obvious that data from referenceable sources is used in Sections B-1 and B-2.

14 Page 17, B-2.4, second paragraph.

The statement that a granite reposttory would be twice the size of a bedded-salt repository should be explained or referenced.

15. Page 18, B-2.5.2, second sentence.

If the leach rate of glass is unimportant, the value of 10-6 g/cm2-day and its "conservativeness" (7-100 times too high or too low?) should be supported, at least by references.

16.

Page 20, B-2.6.

The paragraph on cost considerations is inadequate.

No source is provided for costs. Vague reference is made to sources where data have been developed on comparative costs, but no references are provided. A cost-effectiveness evaluation is recognized, but no support is evidenced. We do not believe the last statement, if we understand it correctly. Section B-4 suffers from the same support deficiency.

17.

Page 23, B-3.2, last line.

Reference is made to (1) " Alternatives for Managing Wastes from Reactors and Past-Fission Operations in LWR Fuel Cycle," ERDA 76-43.

This is a five-volume set.

This and other references to this report should be more specific. There are many other places in the report where references should be made, e.g., first sentence of paragraph 2 on page 24 should reference ERDA-1553-D, Management of Intermediate Level Wastes.

18. Page 25, B-3.2.1, first paragraph, third sentence.

NRC has not accepted either calcine or glass as acceptable

~

forms of HLW.

19.

B-3.2.1.1.1.1, pa ge 28.

Reference is made to " Techniques for Solidification of High Level Wastes," Draft Technical Report, IAEA.

Is this report readily available? Is there assurance that the reference will remain unchanged in the final report?

20. Table B-14, starting on page 29.

No source is provided for this table.

Is is very similar to a table in ERDA 76-43 (ADL reference 1).

Same comment for several other tables.

21. Page 38 B-3.2.1.1.1.1, first paragraph.

A more specific reference should be made to the development and prototype testing program being discussed. Several facilities have been mentioned in the discussion.

22. Page 67, B-3.2.2.1, main paragraph, seventh sentence.

6 Typical hull-waste contamination level (10 mci /g) should be referenced or supported.

23. Page 69, B-3.2.2.2.2.1, second sentence.

The re fo re, aj; men ti on e d.... (i nse rt a s ).

24. Page 81, B-3.2.4, third sentence.

Tritium should be considered as an effluent from solidification, unless special processing at head end is assumed.

This is a minor point.

25. Page 89, B-3.3, second paragraph, last sentence.

Reference to Task A report should be specific. This is a common problem in the reports (A, B, and D at least). Poor referencing or poor methods of derivation do not enhance credibility and make review very difficult.

. 26.

B-3.4, page 96, third paragraph.

This paragraph contains a reference to (16). Reference (16) on page 204 should indicate that 0WI is (or was) with ERDA (00E) not EPA. There are several reports in this series, concerned with shale, granite, basalt, etc., that could be used as references for repository design.

27.

B-3.4.1, page 99.

Discussion concerning incineration and pyrolysis should mention potential radiological impacts from normal and abnormal effluents.

Severe impacts are possible from system upsets, particularly with pyrolysis (destructive distillation) processes.

28. Page 110, B-3.5.1, last paragraph.

Why is the 126-kW/ acre heat load used in this report?

Second, where did the 700-GW addition come from? A previous statement said the repository was sized to receive all waste generated by the LWR fuel cycle through the year 2000.

29.

B-3.5.3.2, paragraph 2, page 137.

This paragraph appears to negate the assumption (on page 17) that a granite repository would be roughly twice the size of a comparable repository in bedded salt.

30.

B-4.2, page 188.

What is the source of these costs? Coment: This same question can be asked (with no apparent answer) in Section B-4 generally.

. References 1.

Letter to Dr. I. B. Wall from David S. Smith, Director, Technology Assessment Division (AW-459) U. S. Environmental Protection Agency, Washington, D.C.

20460, May 20, 1977.

2.

Letter to Dr. Michael Cullingford from John L. Russell, Chief. ESAB (AW-459) U. S. Environmental Protection Agency, Washington, D.C.

20460, December 21, 1977.

3.

Letter to Dr. Michael Cullingford from John L. Russell, Chief.

Energy Systems Analysis Branch, U.S. Environmental Protection Agency, Washington, D.C.

20460, March 21, 1978.

4.

Report Entitled: Technical Support for Waste Management Standards by Arthur D. Little, July,1977.

5.

Letter to Daniel J. Egan, USEPA, from George A. Kiersch, Geologic Consultant, transmitting a report entitled:

" Comments on Geological Aspects Subtask D Report--Assessment of Accidental Pathways."

6.

Draft report entitled: " Critique of ADL Study Part D" by Daniel J.

Kleitman.

7.

Letter to James Stevens, Arthur D. Little inc. from Robert K.

Weatherwax, California Energy Commission, transmitting comments on Subtask D report of Arthur D. Littic, Inc., July 12, 1978.

8.

Letter to Dan Egan, USEPA from J. B. Fussell, JBF Associates, entitled: " Report of Findings concerning ADL Risk Assessment of High-Level Radioactive Waste Disposal, May 19, 1978.

9.

Letter to James Stevens, Arthur D. Little, Inc., from John L. Russell, Chief, Energy Systems Analysis Branch, U. S. Environmental Protection Agency, Washington, D.C.

20460, November 22, 1978.

10.

Slemmons, David B., " Faults and Earthquake Magnitude," State-of-the-art for Assessing Earthquake Hazards in the United States, Report 6 of Miscellaneous Paper S-73-1, MacKay School of Mines, Reno, Nev.,

Microfiche No'. AD-A0 49 870.

11.

D'Appolonia, C nsulting Engineers, Inc., Development of Plan and Approach for Borehole Plugging Field Testing, Report No. ONWT-3, Battelle Memorial Institute, Office of Nuclear Waste Isolation, Col umbus, Ohio,1978.

. 12. Lambe, T. W., Predictions in Scil Engineering: Geotechnique, V. 23, No. 2, pp.149-202,1973.

13.

Interagency Review Group on Nuclear Waste Management, Subgroup Report on Alternative Technology Strategies for tiie Isolation of Nuclear Waste (Draft), TID 28818, 1978.

14.

Gera, Ferruccio, " Review of Salt Tectonics in Relation to-the Disposal of Radioactive Wastes in Salt Formations," Geological Society of America, Bulletin, Vol. 83, No.12, pp. 3551-3574,1972.

15. Machette, Michael N., Dating Quaternary Faults in the Southwestern United States by Using Buried Calcic Paleosols: Jour. Research, U.S. Geological Survey, V. 6, No. 3, pp. 369-381.

16.

" Code of Stratigraphic Nomenclature" prepared by the American Commission on Stratigraphic Nomenr.lature, American Association of Petroleum Geologists, Bulletin, Vol. 45, No. 5, pp. 6451-6665,1961, in conjunction with a correction published in Vol. 45, No. 6, p.1601, 1961.

17. Levorsen, A.

I., Geoloay of Petroleum: A Series of Books in Geology.

W. H. Freeman & Co., 2nd edition,1967.

18.

Inhaber, Herbert, " Risk of Energy Production," Atomic Energy Control Board, No.1119, P. 30, March 1978.

19. Okrent, D., Whipple, C., An Approach to Societal Risk Acceptance Criteria and Risk Management, UCLA-ENG-7746, June 1977.

20.

Dames and Moore, Vol. 21. " Groundwater Movement and Nuclide Transport," Technical Suoport for GEIS: Radioactive Waste Isolation in Geologic Formations, Y/0WI/TM-36/21, Office of Waste Isolation, Oak Ridge, Tennessee,1978, pp. 8-4 through 8-7.

. APPENDIX Table of Contents Pace 1.

Material Supplied to Review Group a.

Objecti ve o f the Revi ew.......................

43 b.

Te n ta ti ve A g e n d a.............................. 44 c.

A t te n de e L i s t................................. 47 d.

Repres entati ve Ques tions...................... 48 2.

Letter from Newell J. Trask to Daniel J. Egan, Follow-up Letter to Review Meeting..............

49

-4 3-Material Supplied to Review Group Review of Work on Risk Assessment of Radioactive Waste Isolation in Deep Geologic Formations Objective - For the purpose of judging its technical validity and usefulness for the support of standards, a review of the Arthur D. Little, Inc. (ADL) report " Technical Support for Radiation Standards for High-Level Radioactive Waste Management" will be performed, Specifically, the following objectives will be fulfilled:

(1) Determine the technical validity of the ADL work by examining the applicability of their models to the analysis performed and the soundness of the data.

(2) Examine the manner in which the ADL work was utilized to formulate the EPA draft standards and the degree to which the ADL results support the standards.

Approach - The persons involved in the review will meet in Bethesda, Maryland, at theluclear Regulatory Commission offices in the Maryland National Bank Building during the week of December 4-8, 1978. The individuals will be assigned to a small working group with a charter to examine specific aspects of the ADL report. The attachment gives a breakdown of the working groups.

The chairman of the group will be responsible for coordinating the activity and producing the report of the review.

NOTE: The ADL dose to man work will not be considered.

Tentative Agenda Technical Review Group Meeting December 4 through Decemoer 8, 1978 Monday, December 4, 1978 9:00 a.m.

Introduction and C. Roberts (Room 6507 MNBB)

Goals of Group (Chairman)

C. Roberts will review the objectives (see Enclosure 1) with the working members.

9:15 a.m.

Overview of ADL Work D. Egan and EPA Thinking on (EPA)

Standards Formulation D. Egan will present EPA thinking on st4ndards formulation and indicate to what extent the ADL report was used in EPA s tandards formulation.

10:30 a.m.

RG Discussion on Task Breakdown and Volunteer Assignment to Subgroup Discussion will address:

1.

The importance that each Review Group member obtain an understanding of how the specific parts he is reviewing according to his specific expertise fits into the overall analysis used to support the EPA standards. Consider how sensitive the final analysis of health effects (or whatever specific terms the EPA standards are given in) is to the separate or component analyses or particular physical para-meters.

2.

The most effective and expeditious way to document the findings of the Review Group. Suggestions are:

focus comments to address only the two major meeting o

objectives (see enclosure 1) focus comments to Working Group subject areas only o

collect comments at end of each daily meeting o

o collect comments only at end of week

1:00 p.m.

RG Members meet in Subgroups to Formulate Plan of Action (in Various Offices in MNBB)

Note:

RG members will be asked to study in working groups of 2-5 people in assigned offices in MNBB.

Tuesday, December 5,1978 8:00 a.m.

Subgroups report RG on Preliminary Insights 9:00 a.m.

Subgroup Working Sessions RG 12:00 Working Luncheon with Subgroup Chairmen (Room 6507 MNBB) 2:00 p.m.

Subgroup Working Sessions RG Wednesday, December 6,1978 8:30 a.m.

Progress Report RG (Room 6507 MNBB) 10:00 a.m.

Questioning of EPA and RG ADL Technical Staff AOL by RG EPA 1:00 p.m.

Subgroup Working Sessions RG Thursday, December 7, 1978 8:30 a.m.

Subgroups Finalize RG Comments 3:00 p.m.

Hand in Draft Comments RG and Circulation

Friday, December 8, 1978 8:30 a.m.

Preparation of Final RG Comments, Discussion 1:00 p.m.

Chairman's Overview 3:00 p.m.

Conclude

- 4 7-ATTENDEE LIST, TECHNICAL REVIEN GROUP MEETING December 4-8, 1978 I.C. Roberts (Chairman)

L.L. Beratan NRC R.R. Bovie R.J. Budnitz

• J.M. C ldwell a

• L.A. Casey E.F. Conti M.C. Cullingford

• J.J. Cu rry

• D.J. Fehrincer J.W. Hickey

• R.B. Hofman n C.P. Jupiter

• M.E. Krug J.C. Palaro P.E. McGrath

• T.J. Nicholson

• E. O' Donnell

• C.B. Oh

• W.R. Pearson

• D.S. Rubinstein

• S.F. Schreurs

• W.E. Vesely

• G.E. Apostolakis UCLA

• J.E. Campbell SLA

• G.D. DeBuchannane USGS

• F.A. Donath UI

• J.C. Hel ton ASU

• C.D. Hollister Woods Hole

• R.L. Iman SLA

• M. Johnson LASL

• F.J. Pearson USGS

• N.J. Trask USGS R.B. Lantz Intera (Informational Consultant)

• Subgroup nembers

.. Some Representative Questions and Areas Which Should Be Included in the Review Technical Review Group Meeting December 4-8, 1978 1.

Are the Models Realistic?

a.

Validity of Assumptions b.

Impact on the analysis results of incorrect assumptions c.

How should identified weaknesses in the models be improved?

2.

Is the data valid?

a.

Estimate the uncertainties in the data used b.

Estimate each data uncertainty and its contribution to the uncertainty in the results 3.

Is the time period (for the standard i.e.,104 yrs) appropriate?

4.

Do the event sequences chosen for calculation cover a reasonably complete range?

a.

Which important potential risk contributors were omitted?

b.

Were the criteria for choice of sequences valid?; quantitative or qualitative?

5.

Were key parameters, processes and events identified?

a.

By a proper sensitivity analysis? or b.

Only by parametric variation?

6.

Were the uncertainties in the results considered?

a.

Were these uncertainties propagated and quar.tified?

b.

By acceptable nun.arical methods?

c.

Were the contributing uncertainties correctly assessed?

7.

Conclusions 8.

Does the ADL report or other reference contain the analysis that leads to your conclusion?

NOTE:

It is important that wherever possible each. specific comment be accompanied by cor nt be an estimate of its contribution to the error in the results and by a specific suggestion for improving the analysis.

49 (P'h United States Department of the Interior M '-

1 h*h..,I) 959 National Center

' _ ev_

GEOLOGIC.\\L SURVEY RESTON, VIRGINIA 22092 n..'

December 19, 1978 Mr. Daniel J. Egan Office Radiation Program (AW-459)

U.S. Environmene.11 Protection Agency Washington, DC 20460

Dear Mr. Egan:

I promised at the NRC revicw of the A.D. Little work on risk assessment to provide some information on volcanism and tectonism that might more accurately reflect current 1cycis of understanding than than presented at the review. Within the constraints of one week and other commitment..;, I have worked up the enclose'd coi:.monts. Like Coorge Debuchannanc. I icci the contractor has the primary obligation to do this sort of thing and is deficient if he doesn't.

However, L hope the comments will be helpful.

Sincerely, f

Newell J. Trask Geologist Enclosure e

e

, Volcanism Ccology may not be basically a predictive science, but one thing we can predict with some certainty is that chains of andesitic strato-volcanoes will occur in the future along active place margins. These areas can be reasonably excluded both from the numerator and denominator of the equation on p. 118. Thst 1 caves us with intraplace volcanism -- much less well understood with regard to space and time relations.

There are two fundamental types of intraplate volcanics -- 1) basic, forced by partial melting in the mantic or lower crust and rising to the surface through narrow pipe and flusures; and 2) silicic, crupted from high-level storage chambers some 10's of Km's in diameter probably in the upper km of the crust. The basic volcanics represent only a minor hazard. A better picture of their formation is emerging as new materials are dated. A compilation of the latest dates will be out shartly as part of the USGS geothermal program. On the basis of data available to me, I esti= ate a high rate of 3x10-3/yr for basic volcanics. If we follow the equation on p. 118, I'

-3

-8 1-3x10

= 10 6

4x10 f

.1 wheretheareadsforthewesternU.S.only. This is close to the value on p. 118 but comes from completely different data.

The apparent rate fc silicic volcanism is much lower, and it may in fact be possible to select a site with essentially 0 probability over 104 years. There have been 15 silicic centers developed on the stable plate in the last 106 years (USGS Circular 726). A high rate assumes that a new one could form anywhere in the western U.S.

Then

- y,600 (1.-5x10-5) 9 6

4x10 2

where the area of influence is H (14)2 ~ 600 Km. Note that for these centers f = 1.

Tectonics The tectonic problem is much more difficult, there being hardly any actualistic codels to draw on.

The discussion on pages 126 to 130 of the A.D. Little report is excellent in pointing up these many diffier: tics. The derivation on p. 136, however, seems highly arbitrary, even na./r,

d4

~.

. 2 In all the history of geologic science, only 3 new faults have been observed and 2 of these were along active place margins.

This suggests that for a repository, the main hazard will be from renewed movement on pre-existing faults. For a repository in sedimentary recks, that includes faults which may not be readily visible, in the underlying basecent.

Recurrence rates on faults are gradually being compiled. The USGS young fault =ap (MF 916) shows the exerc=c variability of recency of movement for tha U.S. and why a generic recurrence rate is almost meaningless. llow can we possibly say what the probability of movement is on some buried fault in the midcontinent?

Perhaps Ve can put an upper bound on recurrence rates. Along the active plate margin in the west the rate is 100-200 vesis; in the Basin and Range province 1000 to 2000 years; presu= ably many faults else-where have much lower rates; many must be essentially O.

The Ramapo fault in New York and New Jersey is insr.ructive in indicating how a fault thought to be inactive may prove to be active.

F.ecurrence rates of about 1000 years seem Indicated by Aggarual and Sykes (1978) on the basic of seismic data.

I assume that we are interested in any movement on the fault, even low intensity events, since these would destroy any grouting and might_ open up a pathway for migrating fluids.

The picture is enortnously complicated by evidence both in the geologic record and historical evidence (going back as cuch as 2000 years in China) that recurrence rates are not constant but change drastically as stress regimes and tectonic provinces migrate. The L= plication of constancy in the ADL report is outdated and naive.

1f we cust have an absolute upper bound on recurrence rates. 10-* per year would seem to be the best vc can do at the present time.

This assumes a Ramapo type fault with a secular rate lower than the present rate on that fault. Certainly the rate will be much lower on many faults, but how we tell which ones aad where is not clear at present.

e 9

O e